JP2993344B2 - Waveform generation device, waveform storage device, waveform generation bubble, and waveform storage method - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a waveform generation device having a waveform memory storing waveform sample data and a waveform storage device (waveform memory) used in the waveform generation device.

[0002]

2. Description of the Related Art Conventionally, as a method of generating a musical tone waveform used in an electronic musical instrument or the like, a waveform amplitude value at a predetermined sampling point of a predetermined waveform is digitally stored in a waveform memory, and the stored contents of the waveform memory are repeated. There is known a waveform generation method for generating a musical tone waveform by reading the waveform.

Reading of waveform data from a waveform memory is performed by
Specifically, it is performed as follows. First, frequency information (so-called F number) F corresponding to the pitch of a musical tone to be generated is sequentially accumulated by an accumulator, and an accumulated value q sequentially outputted
The waveform data in the waveform memory is read at a predetermined cycle using the integer part I of F (q = 1, 2, 3,...) As an address. Moreover, since the waveform data read out using only the integer part I of the accumulated value qF as an address contains quantization noise, this is improved by using an interpolation method. That is, waveform data at a plurality of adjacent sample points is read out, interpolation is performed based on the decimal part d of the accumulated value qF, and an amplitude value at an arbitrary position is obtained and output as waveform data.

In this way, a musical tone waveform having a desired pitch is output. For the waveform data in the waveform memory, the waveform amplitude values at the sample points are sequentially changed as they are (so-called linear).
Stored. Since the value of the F number increases as the pitch increases, the integer part I of the accumulated value qF also increases, and it may be necessary to skip the data in the waveform memory. Since the waveform memory stores the waveform data linearly, the correct waveform amplitude value at each sample point can always be read even if the address is skipped.

On the other hand, for the purpose of suppressing the capacity of the waveform memory, the waveform data is compressed. As a compression method, for example, DPCM (differential PC
M), LPC (Linear Predictive Coding), and ADPCM. Each of these methods is a method of obtaining current waveform data (reproduced value) using past waveform data (reproduced value) and newly read data.

[0006]

However, when the above-described data compression is performed, the waveform data at that time is always calculated by using the data before a certain time, so that it is stored in the waveform memory. Data must be read continuously. If at least one piece of data is not read, it will not be possible to reproduce the next data. Therefore, there is a problem that the waveform memory cannot be skipped and read in order to reproduce a musical tone waveform signal having a high pitch, and the width of the pitch is limited.

The present invention relates to a waveform generation device having a waveform memory storing waveform sample data, and an improvement of a waveform storage device (waveform memory) used in such a waveform generation device.

Further, the present invention provides a waveform generator and a waveform storage device which can skip and read the waveform memory without limiting the pitch width, and can compress the data in the waveform memory. The purpose is to:

[0009]

In order to achieve this object, the invention according to claim 1 provides uncompressed waveform amplitude value data for each of a plurality of predetermined sample points, and sets sample points of the uncompressed waveform amplitude value data. For the sample points in between, the uncompressed waveform amplitude value data can be calculated using the uncompressed waveform amplitude value data of the immediately preceding sample point,
And a waveform storage unit provided with compressed compressed waveform amplitude value data so that uncompressed waveform amplitude value data can be calculated using the uncompressed waveform amplitude value data of the immediately following sample point; Reading means for reading one non-compressed waveform amplitude value data and a plurality of compressed waveform amplitude data included in the read range, and using the read one non-compressed waveform amplitude value data to read the read compressed waveform amplitude value. Calculation means for calculating uncompressed waveform amplitude value data from data
Wherein the one uncompressed waveform amplitude value data read out
Corresponding to sample points on the future side in time
For waveform amplitude data, the uncompressed waveform amplitude data
Compression wave in the direction from the data to the future side (positive direction)
Shape data and calculate the read one uncompressed wave
Sampling point on the past side in terms of shape amplitude value data
The compressed waveform amplitude value data corresponding to
Direction from the compressed waveform amplitude value data to the past (reverse direction)
And a reproducing means for reproducing the waveform data based on the obtained non-compressed waveform amplitude value data.

According to a second aspect of the present invention , the non-compressed waveform amplitude value data is provided for each of a plurality of predetermined sample points, and a sample point between the sample points of the non-compressed waveform amplitude value data is provided. Uncompressed waveform amplitude value data can be calculated using the uncompressed waveform amplitude value data at the immediately preceding sample point, and uncompressed waveform amplitude value data can be calculated using the uncompressed waveform amplitude value data at the immediately following sample point. Thus, a waveform storage means provided with compressed compressed waveform amplitude value data, frequency information generation means for generating frequency information corresponding to the pitch of a musical tone waveform to be generated, and the frequency information repeated and accumulated at a predetermined speed. Accumulating means for calculating and outputting an accumulated value; and
Reading means for reading one piece of non-compressed waveform amplitude data and a plurality of pieces of compressed waveform amplitude data included in a continuous range; and using the one piece of read non-compressed waveform amplitude data to read the compressed waveform amplitude. a calculating means for calculating a non-compressed waveform amplitude value data from the value data, the read
From the viewpoint of one uncompressed waveform amplitude value data
The compressed waveform amplitude value data corresponding to the sample point
From the uncompressed waveform amplitude value data to the future side.
The compressed waveform amplitude value data is sequentially calculated in the
From the output of one uncompressed waveform amplitude value data
Compressed waveform amplitude value data corresponding to past sample points
From the uncompressed waveform amplitude data
To calculate the compressed waveform amplitude value data in the direction toward
Performed for a, from a non-compressed waveform amplitude value data obtained, selecting means for selecting a non-compressed waveform amplitude value data used for the interpolation, an interpolation process using a non-compressed waveform amplitude value data selected by said selection means And interpolation means for outputting the interpolation result as waveform data.

The reading means may determine a read address of the waveform storage means based on an upper value of the accumulated value. In the embodiment described later, the read address of the waveform memory is determined based on the integer part of the accumulated value qF.

[0012]

Further, the waveform storage device according to claim 3 is
The sound was obtained by sampling at a predetermined sampling frequency
A waveform storage device for compressing and storing waveform data,
Obtaining n samples of the waveform data, providing M (M is an integer of M ≦ n / 2 + 1) uncompressed waveform amplitude value data for each of a plurality of predetermined sample points, For the sample points between the sample points, the uncompressed waveform amplitude value data at the immediately preceding sample point can be calculated using the uncompressed waveform amplitude value data at the immediately preceding sample point. Compressed data so that uncompressed waveform amplitude value data can be calculated using the data
(N-1) pieces of compressed waveform amplitude value data are provided. According to a fourth aspect of the present invention, in the third aspect, one uncompressed waveform amplitude value data is arbitrarily selected from the uncompressed waveform amplitude value data provided for each of the plurality of predetermined sample points. Using the uncompressed waveform amplitude value data, when calculating a plurality of corresponding non-compressed waveform amplitude value data from a plurality of continuous compressed waveform amplitude value data immediately after the uncompressed waveform amplitude value data, The last non-compressed waveform amplitude value data of the non-compressed waveform amplitude value data is the same as the non-compressed waveform amplitude value data provided next to the selected one non-compressed waveform amplitude value data. The invention according to claim 5 is a waveform generating method for reading and reproducing waveform data from the waveform storage means, wherein the waveform storage means provides uncompressed waveform amplitude value data for each of a plurality of predetermined sample points, For the sample points between the sample points of the uncompressed waveform amplitude value data, the uncompressed waveform amplitude value data at the immediately preceding sample point can be used to calculate the uncompressed waveform amplitude value data, and In order to be able to calculate the non-compressed waveform amplitude value data using the non-compressed waveform amplitude value data at the points, the compressed compressed waveform amplitude value data is provided, and the waveform data is read and reproduced from the waveform storage means. A reading step for reading one uncompressed waveform amplitude value data and a plurality of compressed waveform amplitude value data included in a continuous range from said waveform storage means. When was the read 1
Calculating uncompressed waveform amplitude value data from the read compressed waveform amplitude value data using the two uncompressed waveform amplitude value data, and calculating time from the read one uncompressed waveform amplitude value data. As for the compressed waveform amplitude value data corresponding to the sample point on the future side, the compressed waveform amplitude value data is sequentially calculated in the direction toward the future side from the uncompressed waveform amplitude value data, and the one read non-compressed waveform amplitude value data is calculated. Regarding the compressed waveform amplitude data corresponding to the sample point on the past side in terms of the compressed waveform amplitude data, the compressed waveform amplitude value data is sequentially transferred from the uncompressed waveform amplitude data in the direction toward the past. A calculating step for calculating;
A reproducing step of reproducing waveform data based on the obtained uncompressed waveform amplitude value data. The invention according to claim 6 is a waveform generating method for reading out and reproducing waveform data from the waveform storage means, wherein the waveform storage means provides uncompressed waveform amplitude value data for each of a plurality of predetermined sample points. For the sample points between the sample points of the compressed waveform amplitude value data, the uncompressed waveform amplitude value data of the immediately preceding sample point can be used to calculate the uncompressed waveform amplitude value data, and the immediately following sample point In order to calculate the non-compressed waveform amplitude value data using the uncompressed waveform amplitude value data, the compressed waveform amplitude value data is provided, and reading and reproduction of the waveform data from the waveform storage means is performed. A frequency information generating step of generating frequency information corresponding to the pitch of a musical tone waveform to be generated; and repeatedly accumulating the frequency information at a predetermined speed to output an accumulated value. An accumulating step, and a reading step of reading one uncompressed waveform amplitude value data and a plurality of compressed waveform amplitude data included in a continuous range from the waveform storage means based on the accumulated value. A calculating step of calculating non-compressed waveform amplitude value data from the read compressed waveform amplitude value data using one read non-compressed waveform amplitude value data; As for the compressed waveform amplitude value data corresponding to the sample point on the future side in view, the compressed waveform amplitude value data is sequentially calculated in the direction toward the future side from the uncompressed waveform amplitude value data, and the readout is performed. With respect to the compressed waveform amplitude data corresponding to the sample point on the past side in terms of one non-compressed waveform amplitude value data, A calculating step of sequentially calculating compressed waveform amplitude value data in a direction toward the other side; a selecting step of selecting non-compressed waveform amplitude value data to be used for interpolation from the obtained uncompressed waveform amplitude value data; And performing an interpolation process using the non-compressed waveform amplitude value data selected in step (a), and outputting an interpolation result as waveform data. Furthermore waveform storage method according to claim 7, sound
Wave obtained by sampling at a predetermined sampling frequency
A waveform storage method for compressing and storing shape data.
After obtaining n samples of the waveform data , M (M is an integer of M ≦ n / 2 + 1) uncompressed waveform amplitude value data is provided for each of a plurality of predetermined sample points, and For the sample points between the sample points, the uncompressed waveform amplitude value data at the immediately preceding sample point can be calculated using the uncompressed waveform amplitude value data at the immediately preceding sample point. Compressed data so that uncompressed waveform amplitude value data can be calculated using the data
It is characterized in that (n-1) pieces of compressed waveform amplitude value data are provided. The invention according to claim 8 is the invention according to claim 7, wherein arbitrarily one uncompressed waveform amplitude value data is selected from the uncompressed waveform amplitude value data provided for each of the plurality of predetermined sample points, and the selected non-compressed waveform amplitude value data is selected. When a plurality of uncompressed waveform amplitude value data corresponding to a plurality of continuous compressed waveform amplitude value data immediately after the uncompressed waveform amplitude value data is calculated using the compressed waveform amplitude value data, The last non-compressed waveform amplitude value data of the compressed waveform amplitude value data is the same as the non-compressed waveform amplitude value data provided next to the selected one non-compressed waveform amplitude value data.

[0014]

The waveform storage means stores amplitude value data at a plurality of sequential sample points. The amplitude value data is either uncompressed waveform amplitude value data or compressed waveform amplitude value data. The uncompressed waveform amplitude value data is provided for each of a plurality of predetermined sample points. The compressed waveform amplitude value data is provided between the non-compressed waveform amplitude value data. From the compressed waveform amplitude value data, the non-compressed waveform amplitude value data can be calculated using the immediately preceding non-compressed waveform amplitude value data. Waveform amplitude value data can also be calculated.

When a musical sound waveform is generated, first, frequency information corresponding to the pitch of the musical sound waveform to be generated is generated. This frequency information is repeatedly accumulated at a predetermined speed, and the accumulated value is output. Based on the accumulated value, one uncompressed waveform amplitude value data and a plurality of compressed waveform amplitude value data included in a continuous range are read from the waveform storage means. The read compressed waveform amplitude value data is converted into uncompressed waveform amplitude value data.

From the uncompressed waveform amplitude data obtained as described above, some uncompressed waveform amplitude data to be used for interpolation are selected. An interpolation process is performed using the selected uncompressed waveform amplitude value data, and the interpolation result is output as waveform data.

The data used for interpolation may be selected from the data read from the waveform storage means, and the compressed waveform amplitude data of the selected data may be converted to non-compressed waveform amplitude data.

[0018]

Embodiments of the present invention will be described below with reference to the drawings.

FIG. 1 shows a block configuration of an electronic musical instrument to which a waveform generating device according to an embodiment of the present invention is applied. This electronic musical instrument includes a panel switch (SW) 1, a keyboard 2, a microcomputer (microcomputer) 3, a microphone 4,
Analog-to-digital converter (ADC) 5, buffer RA
An M (random access memory) 6, a tone generator 7, a waveform memory 8, a digital-to-analog converter (DAC) 9, and a sound system 10 are provided.

The panel switch 1 is a switch provided on the panel, such as a tone switch for designating a tone of a musical tone to be generated. Operation information output in response to the operation of the panel switch 1 is input to the microcomputer 3. The keyboard 2 is a keyboard provided with a plurality of keys for performing by a player. Performance information (such as a note on / off signal indicating a key on / off and a note code indicating a pitch) output from the keyboard 2 according to the performance of the player is input to the microcomputer 3.

The tone generator 7 reads out waveform data from the waveform memory 8 in response to an instruction from the microcomputer 3, and generates and outputs a tone signal (digital waveform amplitude value data). The sound source unit 7 will be described later in detail.

The waveform memory 8 stores the waveform data R
AM and ROM (read only memory).
The RAM area of the waveform memory 8 stores externally input waveform data. The ROM area of the waveform memory 8 stores preset waveform data set in advance.

The microphone 4 is for recording external sounds. An external sound is input as an analog signal by the microphone 4 and is sampled at a predetermined sampling frequency by the ADC 5 to be a digital signal. AD
The digital signal output from C5 is compressed by the microcomputer 3 into a format shown in FIG. 2 described later, and stored in the RAM area of the waveform memory 8 via the sound source unit 7. The buffer RAM 6 is a buffer used when the microcomputer 3 compresses the waveform data.

The DAC 9 performs digital-to-analog conversion on the digital tone signal from the tone generator 7 and outputs an analog tone signal. The sound system 10
A tone is actually emitted based on the analog tone signal output from C9.

The microcomputer 3 controls the operation of the entire electronic musical instrument of this embodiment. In particular, when recording an external sound in the waveform memory 8, a process of compressing the digital signal from the ADC 5 into a format shown in FIG. This will be described later with reference to FIG. When a key on the keyboard 2 is pressed, the microcomputer 3 inputs performance information output from the keyboard 2 in response to the pressing, and instructs the sound source unit 7 to sound according to the performance information.

FIG. 2 shows the format of waveform data in the waveform memory (RAM area and ROM area) 8 of FIG. The waveform memory 8 has a plurality of storage banks, and each bank stores waveform data for each tone color. FIG.
This shows the leading part of waveform data of a certain tone color.

The waveform data is composed of successive sample point amplitude value (waveform amplitude value) data of the musical tone waveform of the timbre. However, in this embodiment, uncompressed data (hereinafter, referred to as “linear sample”) of the amplitude value itself and difference data (hereinafter, “compressed sample”) from the amplitude value of the immediately preceding sample point are used.
), And the difference data is also a difference from the amplitude value of the immediately following sample point.

In FIG. 2, the address-
.., 2, -1, 0, 1, 2, 3,... Indicate offset addresses based on the position of the start address. Sound source section 7
When the waveform memory 8 is actually accessed inside the device, since the access is performed based on the position of the linear sample, the position of the start address is set to the position of the first linear sample appearing for convenience of explanation, Compressed samples corresponding to the addresses (address-1, -2) are stored.

L is a 16-bit linear sample, D is 8
Indicates a compressed sample of bits. In the area of one address, one data can be stored for a linear sample, and two data can be stored for a compressed sample. Sound source section 7
When the waveform data of the waveform memory 8 is read from the memory, the data is read in the above-mentioned address unit. That is, by specifying and reading one address, it is possible to read one data of a 16-bit linear sample for a linear sample and two data of an 8-bit compressed sample for a compressed sample.

In the figure, the parenthesized subscripts attached to L and D are the numbers of sample points (sampling points) corresponding to the data (linear samples or compressed samples). That is, L (0), L (6), L (12),.
Are the amplitude values (linear sample values) of the 0th sample point, the 6th sample point, the 12th sample point,.
(1), D (2), D (3),... Are the immediately preceding sample point amplitude values (linear sample values of the immediately preceding sample point) of the first sample point, the second sample point, the third sample point,.
Is the difference from.

Specifically, the linear sample value L (1) at the first sample point is obtained from the linear sample value L (0) at the immediately preceding sample point 0 and the difference D (1) by L (1) = L
It can be calculated by (0) + D (1). The linear sample value L (2) at the second sample point is calculated from the linear sample value L (1) at the immediately preceding first sample point and the difference D (2) by L
(2) = L (1) + D (2) Similarly, the compressed sample D (m) and the immediately preceding linear sample L
Using (m-1), a linear sample L (m) = L
(M-1) + D (m) can be calculated.

The waveform data of the present embodiment is configured by sequentially arranging one linear sample in an area for one address and six compressed samples in an area for three addresses.
In particular, a linear sample that can be calculated from the last compressed sample of the six compressed samples is stored as the next linear sample. For example, using the compressed sample D (6) at address 3, L (6) = L (5) +
D (6), and its linear sample L (6)
It is stored at the next address 4.

Therefore, the compressed sample D is a difference from the amplitude value (linear sample value) of the immediately preceding sample point and also a difference from the amplitude value (linear sample value) of the immediately following sample point. That is, in the above description, the linear samples are sequentially calculated as L (m) = L (m-1) + D (m) in the direction in which the address increases (positive direction), but in the reverse direction (the direction in which the address decreases). Linear samples can also be calculated.

For example, the linear sample L of address 4
Using (6) and the compressed sample D (6) at address 3, the linear sample L (5) at the fifth sample point is
(5) = L (6) -D (6) Using the linear sample L (5) and the compressed sample D (5), a linear sample L (4) = L at the fourth sample point is obtained.
(5) -D (5) can be calculated. In the same manner, a linear sample can be calculated in the opposite direction by L (m) = L (m + 1) -D (m + 1). Therefore, each compressed sample D
(M) can also be said to be a difference from the amplitude value of the sample point immediately after the (m-1) th sample point (the linear sample value of the sample point immediately after).

In this embodiment, the L of the linear sample
Since the position of the sample point is specified with reference to the position of (0), if the sample points that can be reproduced from the waveform data in FIG. 2 are described from the beginning, the following is obtained. For convenience of explanation, it is assumed that the sampling points are specified by the following designations.

Fourth sample point: Linear sample is L (-4) Third sample point: Linear sample is L (-3) Second sample point: Linear sample is L (-2) First sample point ... Linear sample is L (-1) 0th sample point ... Linear sample is L (0) 1st sample point ... Linear sample is L (1) 2nd sample point ... Linear sample is L (2) 3rd sample point ... The linear sample is L (3)::

Of the above, the linear sample at the sample point (1) is obtained in the reverse direction from the linear sample L (0) at address 0. The linear samples at the subsequent sample points can be obtained in the forward direction or in the reverse direction from the linear samples held in the waveform data.

The compressed samples D (-3) and D (-
1), D (1), D (3),... Are stored in the lower 8 bits of the 16-bit area of each address.
(-2), D (0), D (2), D (4),... Are stored in the upper 8 bits of the 16-bit area of each address.

FIG. 3 shows how write data is generated when writing waveform data to the RAM area of the waveform memory 8.

An analog tone signal input by the microphone 4 in FIG. 1 is converted by the ADC 5 into a digital tone signal. L (−4), L (−3), L (−) in FIG.
2),... Are linear waveform data L output from the ADC 5, and are linear sample values (16 bits) at each sample point (also described in FIG. 2). The address given below the rectangular block indicating the linear sample is a relative offset address based on the position of the start address, as in FIG.

The linear waveform data L is sequentially input to the microcomputer 3 from the ADC 5 in FIG. The microcomputer 3 generates difference waveform data D from the linear waveform data L as shown in FIG. Specifically, the address m (m = −3, −
2, (1,1,0,1,2, ...), D (m) = L
(M) -L (m-1) and each differential waveform data (8 bits)
Generate Further, two differential waveform data (for example,
D (-3) and D (-2), D (-1) and D (0), D
(1) and D (2)) to generate 16-bit write data.

As described in FIG. 2, D (-3), D (-3)
(-1), D (1), D (3),...
D (-2), D (0), D (2), D (4),... The microcomputer 3 is a buffer RAM of FIG.
6 is used as a work area to generate such write data to the waveform memory 8.

Data to be finally written from the microcomputer 3 to the waveform memory 8 via the sound source unit 7 includes every six linear samples L (0), L (6), L (12),. This is differential waveform data. As a result, the waveform data in the format as described with reference to FIG.

FIG. 4 is a diagram for explaining the principle of reading waveform data in this embodiment.

FIG. 4A shows the waveform memory 8 explained in FIG.
FIG. 4B shows a state in which the waveform data is read in the forward direction (the direction in which the address increases), and FIG. 4B shows a state in which the waveform data is read in the reverse direction (the direction in which the address decreases). Reference numerals 401 and 402 indicate musical tone waveforms to be reproduced. The tone waveforms 401 and 402 have the same waveform. It is a diagram for explaining the principle, and although the reference vertical axis and horizontal axis are not shown, the vertical axis represents the waveform amplitude value, and the horizontal axis represents time (from left to right in the figure, the direction in which time passes). ).

Black and white circles on the tone waveforms 401 and 402 indicate sample points (sampling points).
L (0), L (1), L near the white or black circle
(2),... Indicate linear samples at each sample point.
In particular, white circles indicate sample points where linear samples are held in the waveform data of FIG. 2, and black circles indicate sample points where compressed samples are held instead of linear samples.

FIG. 4 (a) shows a case where the difference data is added to the linear sample and the immediately following linear sample is obtained when the waveform is reproduced in the forward direction. For example, L (1) is calculated by L (0) + D (1),
(2) is calculated by L (1) + D (2), and so on.

On the other hand, FIG. 4 (b) shows a case where the difference data is subtracted from the linear sample and the immediately preceding linear sample is obtained when the waveform is reproduced in the reverse direction. For example, L (5) is calculated by L (6) -D (6), L (4) is calculated by L (5) -D (5), and so on.

In the present embodiment, the tone generator 7 performs time-division sound processing for a plurality of channels. Since the reading process of the waveform data in each channel is the same, the following description will be made focusing on one channel.

The tone generator 7 of this embodiment calculates and outputs one waveform amplitude value by performing an interpolation process (four-point interpolation) using four adjacent linear samples. In order to obtain four linear samples, the tone generator 7 reads data of three adjacent addresses at a time from the waveform data of FIG.
There are the following three types of methods for reading the waveform data of three adjacent addresses including linear samples.

(Type 1) In this case, after a linear sample is read, two compressed samples at the address immediately before the linear sample are read, and two compressed samples at the address immediately before the linear sample are further read. For example, address 4 (linear sample L (6)), address 3 (compressed sample D (5), D (6)) and address 2 (compressed sample D (3), D (4)) in FIG. This is the case. From these read data, as described in FIG. 2 and as can be seen from FIG.
Four linear samples L (5), L in the reverse direction from (6)
(4), L (3), L (2) can be calculated, and as a result, five linear samples L (2) to L (6) can be obtained.

(Type 2) In this case, after reading a linear sample, two compressed samples before and after the linear sample are read. For example, address 4 (linear sample L (6)) and address 3 (compressed sample D
(5), D (6)) and address 5 (compressed samples D (7), D (8)). From these read data, two linear samples L (5) and L (4) can be calculated in the reverse direction from the linear sample L (6), and two linear samples in the forward direction can be calculated from the linear sample L (6). Samples L (7) and L (8) can be calculated. As a result, the linear samples L (4) to L (4)
(5) can be obtained.

(Type 3) In this case, after a linear sample is read, two compressed samples at the next address of the linear sample are read, and two compressed samples at the next address are read. For example, address 4 (linear sample L (6)) and address 5 (compressed sample D
(7), D (8)) and address 6 (compressed samples D (9), D (10)). From these read data, four linear samples L (7), L (8), L
(9), L (10) can be calculated. As a result, five linear samples L (6) to L (10) can be obtained.

Arrows 411, 412, and 413 in FIG. 4 indicate the range of linear samples that can be calculated using data for three addresses including linear samples in the waveform data. For example, in arrow 412, linear sample L
The waveform data for three addresses including (6) indicates that the linear samples L (2) to L (10) can be reproduced.

The waveform data shown in FIG.
Since linear samples are held for every four addresses as in..., Considering the range of linear samples that can be reproduced by the waveform data for three addresses including each of the linear samples, the adjacent ranges are three. It will have an overlap of sample points. For example, in the range of the arrow 411 and the range of the arrow 412, L (2), L
The three points (3) and L (4) overlap, and L (8), L (9), and L (10) in the range of arrow 412 and the range of arrow 413.
3 points overlap.

Therefore, it can be seen that the linear samples of any four adjacent points can be calculated by reading only three addresses including the linear samples in the waveform data of FIG. For example, four points L (1), L (2), L (3), L
In order to obtain (4), 3 including the linear sample L (0)
Addresses (addresses 0, 1, 2 in FIG. 2) may be read, and the next four points L (2), L (3), L (4), L (5)
In order to obtain the address, three addresses (addresses 2, 3, and 4 in FIG. 2) including the linear sample L (6) may be read.

As described above, four points necessary for performing four-point interpolation are obtained.
In order to obtain a linear sample of a point, it is sufficient to read three addresses including the linear sample in the waveform data of FIG. 2, thereby making it possible to skip the waveform data.

Next, a detailed description will be given of the tone generator 7 of the present embodiment, which constitutes a waveform memory as described above and enables skip reading.

FIG. 5 shows a block diagram of the tone generator 7 of FIG. The tone generator 7 includes a control register 501, an F-number generator 502, a counter 503, multipliers 504 and 505,
It comprises an adder 506, an auxiliary address generator 507, a decoder 508, a data selector 509, an interpolator 510, an envelope generator 511, a multiplier 512, and a channel (CH) accumulator 513. Decoder section 50
8, the data selection unit 509, and the interpolation unit 510 are referred to as a reproduction unit.

The tone generator 7 emits a plurality of channels (16 channels / DAC) in a time-division manner. Specifically, the processing of the channel is performed in the time slot of each channel. The processing of each channel is independent. The following description focuses on processing in one channel.

The control register 501 corresponds to the microcomputer 3 shown in FIG.
Various kinds of control data transmitted from the to the sound source unit 7 are stored. Control data input from the microcomputer 3 to the control register 501 will be described. First, when writing waveform data to the waveform memory 8 (at the time of recording), the write data described in FIG. This write data is written to the waveform memory 8 via the control register 501.

Next, the control data input from the microcomputer 3 to the control register 501 at the time of a sound generation instruction (at the time of reproduction) will be described. When any key is depressed on the keyboard 2 of FIG. 1, a note-on signal indicating a key-on and a note code indicating a pitch are output from the keyboard 2 as performance information. These pieces of performance information are input to the microcomputer 3. The microcomputer 3 allocates sound channels to generate musical tones according to the performance. Then, the assigned channel is specified, and the envelope generator (EG) control data, the start address, the note code, and the note-on signal are output to the control register 501 of the tone generator 7. The control register 501 outputs these pieces of information on the sounding channel to each unit in the sound source unit in the time slot of each sounding channel. As a result, a tone signal generation process for each channel is performed. The EG control data and the start address use data corresponding to the tone color set at that time. The setting of the timbre is based on the panel SW1 in FIG.

The EG control data is data indicating the type of envelope waveform according to the designated tone color.
The EG control data is input to the envelope generator 511 via the control register 501. The start address is the start address (the start address of the data in FIG. 2) in the waveform memory 8 of the waveform data corresponding to the designated tone color. The start address is input to the adder 506.
The note code indicates the pitch of a musical tone to be pronounced. The note code is input to the F number generation unit 502. The note-on signal indicates a key press. The note-on signal is input to the counter 503 and the envelope generator 511 as a signal indicating a sound generation start timing.

The components after the F-number generation unit 502 in the sound source unit 7 will be described.

First, the F number generating section 502 generates an F number corresponding to a note code (pitch) of a musical tone to be generated in a channel in a time slot of each channel. The counter 503 is reset by a note-on signal from the control register 501, receives the F number, and sequentially accumulates the accumulated value qF (q = 0, 1,
2, 3, ...). The accumulation is performed for each time slot of the channel.

In this embodiment, the accumulated value qF output from the counter 503 is 30-bit data. The integer part AI (upper 19 bits) of the accumulated value qF is input to the multiplier 504, and the decimal part AF (lower 11 bits) is inputted to the multiplier 505.
To enter.

The multiplier 504 includes an integer part AI and a constant “4”.
And multiply by The multiplication result (21 bits) is added to the adder 50
Enter 6 The reason why the integer part AI is multiplied by the constant “4” is that the linear sample of the waveform data in FIG. 2 is addressed by the result of adding the result of the multiplication and the start address. As shown in FIG. 2, since the linear sample is held at addresses 0, 4, 8,..., The result of multiplication by a constant “4” is a multiple of 4, and the integer part AI is multiplied by a constant “4”. If the start address is added, the linear sample of the waveform data in FIG. 2 will be addressed.

The multiplier 505 includes a decimal part AF and a constant “3”.
And multiply by The higher two bits AF1 of the multiplication result (13 bits) are input to the auxiliary address generator 507 and the decoder 508. Also, a 3-bit AF2 (that is, the upper 2 bits AF1 of the multiplication result of the multiplier 505 plus the most significant bit of the lower 11 bits of the multiplication result) (ie,
The upper 3 bits of the multiplication result of the multiplier 505 are input to the data selection unit 509 and the interpolation unit 510. Further, the lower 10 bits AF3 of the multiplication result of the multiplier 505 are input to the interpolation unit 510.

The auxiliary address generator 507 includes a multiplier 50
5 generates an auxiliary address based on the data AF1. The generated auxiliary address is input to the adder 506. The adder 506 adds the output of the multiplier 504, the start address, and the auxiliary address. The waveform data is read from the waveform memory 8 using the addition result as an address.

As described above, the linear sample of the waveform data shown in FIG. 2 is addressed by adding the output of the multiplier 504 and the start address, so that the auxiliary address indicates an offset from the position of the linear sample. . Further, in this embodiment, since the waveform data for three addresses is read in order to obtain one waveform amplitude value, the auxiliary address generator 507 outputs three auxiliary addresses sequentially according to the input of AF1. Has become. Especially,
Since the linear sample must be read, the auxiliary address output first is "0".

In accordance with the three auxiliary addresses sequentially output from auxiliary address generator 507, adder 506 sequentially outputs three read addresses. As a result, the waveform data in the waveform memory 8 is sequentially read in three times for three addresses. The read waveform data for three addresses is sequentially input to the decoder unit 508. The waveform data for three addresses includes linear samples for one address and compressed samples for two addresses (a total of four compressed samples). Of these, the linear samples are read out first and input to the decoder unit 508. It has become. This is to facilitate the processing in the decoder unit 508.

The decoder section 508 calculates the value of a linear sample from one input linear sample and four input compressed samples. In total, five linear samples are obtained. The decoder unit 508 sequentially outputs the five linear samples to the data selection unit 509.

The data selection unit 509 includes a decoder unit 508
4 used for interpolation from 5 linear samples input from
A process for selecting one is performed. The selected four linear samples are sequentially input to the interpolation unit 510. Interpolator 510
Performs four-point interpolation using these four linear samples, and outputs the result to the multiplier 512.

On the other hand, the envelope generator 511 outputs an envelope waveform from the rising timing of the note-on signal. The envelope waveform to be output is selected according to the EG control data.

The multiplier 512 gives the interpolation result output from the interpolation section 510 the envelope output from the envelope generator 511. Channel accumulator 513
Accumulates the multiplication result (digital tone signal with an envelope) output from the multiplier 512 in the time slot of each channel, and outputs the result to the DAC 9 in FIG.

The outline of the processing in the sound source section 7 has been described above. Next, a detailed description will be given of a process of reading data of three necessary addresses from the waveform memory 8, calculating four necessary linear samples, and performing interpolation in the tone generator 7.

FIG. 6A is a time chart showing the progress of an address when sequentially reading and reproducing the waveform data of FIG. 2 and four adjacent linear samples obtained from the waveform data read at such an address. It is.

In this figure, “AI”, “AF1”,
The columns of “AF2” and “AF3” indicate the values of the data AI, AF1, AF2, and AF3 described with reference to FIG. 5, respectively. The column of “read address” indicates a read address output from the adder 506. The column of “(auxiliary address)” under “read address” indicates the auxiliary address generator 5
Indicates the value of the auxiliary address output from 07. The column of “Necessary 4 samples” at the bottom is AI, AF1, A
Four linear samples to be obtained from the values of F2, AF3 and the auxiliary address are shown.

Now, pronunciation is instructed on a certain channel,
From the counter 503 in FIG. 5, the accumulated value qF (q = 0, 1,
2, 3,...) Are sequentially output. In particular,
, 0, F, 2F, 3F, 4F,... Are output from the counter 503 for each time slot of the channel. A when the output from the counter 503 sequentially increases from 0
FIG. 6 shows changes in values of I, AF1, AF2, and AF3.
This will be described with reference to FIG.

The output from the counter 503 is an integer part AI
And fractional part AF. The integer part AI is initially “0” (while q is small). While AI = 0, the decimal part AF sequentially increases from 0 and approaches 1. When the output from the counter 503 further increases, the integer part AI =
It becomes 1 and the decimal part AF becomes 0 again. And AI =
During 1, the fractional part AF sequentially increases from 0 and approaches 1. Hereinafter, similarly, the integer part AI and the decimal part AF change.

First, attention is paid to the range of AI = 0. AI =
During 0, the decimal part AF sequentially increases in the range of 0 ≦ AF <1, and as a result of multiplying the decimal part AF by the constant “3” in the multiplier 505, 3 · AF is in the range of 0 ≦ 3 · AF <3. Increases sequentially. Therefore, the upper two bits AF1 of the result of the multiplication by the multiplier 505 take values of 0 at the beginning, 1 at the beginning, and 2 at the beginning when the two bits are regarded as an integer. The column of AF1 in FIG. 6A shows such a change of AF1.

AF2 is a multiplier 5 as described with reference to FIG.
05 are the upper 3 bits of the 13 bits of the output
In other words, it is 3 bits obtained by adding 1 bit immediately below 2 bits of AF1. Therefore, AF2 is 3
Assuming that the bit is an integer, it initially takes a value of 0 and then takes a value of 1 while AF1 = 0, takes a value of 2 and then takes a value of 3 during AF1 = 1, and takes a value of 3 first while AF1 = 2. Takes 4 and then takes the value of 5. The column of AF2 in FIG. 6A shows such a change in AF2.

AF3 is the lower 10 bits of the 13-bit output of the multiplier 505 as described with reference to FIG. This is based on the 13-bit output of the multiplier 505
Since the bit AF2 is excluded, the AF3
In each section in which 2 takes a value from 0 to 5, it changes so as to sequentially increase from 0 to a predetermined value (10 bits are all 1). The column of AF3 in FIG. 6A shows such a change of AF3. In particular, in AF3, the state of sequentially increasing is shown by a sawtooth wave.

In the above description, attention was paid to the range of AI = 0, but as shown in FIG. 6A, AI = 1, 2, 3,
..., but AF1, AF2 and AF3 change similarly.

Next, AI, AF1,
The operation of reading the necessary three addresses of waveform data from the waveform memory 8 according to AF2 and AF3 will be described.

As described with reference to FIG. 5, the linear sample of the waveform data shown in FIG. 2 is addressed based on the result obtained by adding the start address to 4.multidot.AI output from the multiplier 504. The parenthesized description in the column of AI in FIG. 6A shows the linear sample in the waveform data of FIG. 2 addressed at the value of the AI and its address.

For example, in the column of AI = 0, “(0: L
(0)) "indicates that when AI = 0, data of three addresses including the linear sample L (0) at address 0 in the waveform data of FIG. 2 is read. ing. In the column of AI = 1, “(4: L
(6)) "indicates that when AI = 1, data for three addresses including the linear sample L (6) at address 4 in the waveform data in FIG. 2 is read. ing.

When the linear sample to be addressed is determined by adding the output 4 · AI of the multiplier 504 and the start address, the auxiliary address output from the auxiliary address generator 507 is an offset from the linear sample. The data preceding and succeeding by the auxiliary address is read from the linear sample. Actually, the auxiliary address generation unit 507 performs the processing shown in FIG.
An auxiliary address is generated as shown in the column of "(auxiliary address)" in (a). The generated auxiliary address is added to the adder 5
Enter 06. Hereinafter, a description will be given for each value of AF1. It is assumed that the start address is 0.

When AF1 = 0 At this time, 0, -1, and -2 are sequentially output as auxiliary addresses. This is the type 1 reading method described above. That is, a linear sample is firstly read at the auxiliary address 0, then two compressed samples of the immediately preceding address are read, and two compressed samples of the immediately preceding address are further read.

For example, when AI = 0, the output of adder 506 (ie, the read address of waveform memory 8) is
Since it is the result of addition of the start address 0 and 4 · AI 0 and the auxiliary address, they are 0, −1, and −2. Therefore, the waveform data of the read addresses 0, -1, and -2 sequentially output from the adder 506 are read and input to the decoder unit 508 in this order. Similarly, when AI = 1, the read address is 0 and 4 · A of the start address.
Since the result of addition of 4 of I and the above auxiliary address,
4, 3, and 2. Therefore, the waveform data at the read addresses 4, 3, and 2 are read and input to the decoder unit 508 in this order.

When AF1 = 1 At this time, 0, -1, and +1 are sequentially output as auxiliary addresses. This is the type 2 reading method described above. That is, first, a linear sample is read at the auxiliary address 0, then two compressed samples of the immediately preceding address are read, and furthermore, two compressed samples of the next address of the linear sample are read.
One compressed sample is read.

For example, when AI = 0, the read address is 0, -1, 1 because it is the result of adding the start address 0, 4 · AI 0, and the auxiliary address.
Therefore, the waveform data of the read addresses 0, -1, and 1 sequentially output from the adder 506 are read and input to the decoder unit 508 in this order. Similarly, when AI = 1, the read address is 0 and 4 · A of the start address.
Since the result of addition of 4 of I and the above auxiliary address,
4, 3, and 5. Therefore, the waveform data of the read addresses 4, 3, and 5 are read and input to the decoder unit 508 in this order.

When AF1 = 2 In this case, 0, +1, +2 are sequentially output as auxiliary addresses. This is the type 3 reading method described above. That is, a linear sample is first read at the auxiliary address 0, two compressed samples at the next address are read, and two compressed samples at the next address are read.

For example, when AI = 0, the read address is 0, +1, +2 because it is the result of addition of the start address 0, 4 · AI 0 and the auxiliary address. Therefore, the waveform data of the read addresses 0, 1, and 2 sequentially output from the adder 506 are read and input to the decoder unit 508 in this order. Similarly, when AI = 1, the read address is 0 and 4
Since the result of addition of AI 4 and the auxiliary address is
4, 3, and 5. Therefore, the waveform data of the read addresses 4, 3, and 5 are read and input to the decoder unit 508 in this order.

FIG. 6B is a time chart showing a specific output state of the auxiliary address. Reference numeral 601 denotes a time slot for each channel. The time slot of the previous channel, the time slot of the current channel of interest, and the time slot of the next channel are arranged in this order. The time slot of each channel is divided into six time slots, and 0, 1, 2,..., 5 are described in the sections of each time slot. Hereinafter, these time slots are referred to as time slot 0, time slot 1,...

As shown in this figure, when AF1 = 0, the auxiliary address generator 507 sets 0 in time slots 0 and 1, -1 in time slots 2 and 3, and-in time slots 4 and 5. 2 are output. A
When F1 = 1, 0 is set in time slots 0 and 1,
-1 in time slots 2 and 3 and 4 in time slot 4
And +1 are output at 5 and 5, respectively. When AF1 = 2, 0 is set in time slots 0 and 1, +1 is set in time slots 2 and 3, and +1 is set in time slots 4 and 5.
2 is output.

Next, the decoder unit 50 which receives the waveform data for three addresses read out as described above, obtains five linear samples from the input data, and sequentially outputs them.
8 will be described in detail.

FIG. 7 shows a reproducing section, that is, a decoder section 50.
8 shows a detailed block configuration of the data selection unit 509 and the interpolation unit 510. Referring to FIG.
Includes a latch 701, selectors 702 and 703, an adder / subtractor 704, a delay (delay) circuit 705, and a decode controller 706.

The latch 701 latches read data (16 bits) from the waveform memory 8. Selector 702
Has three input terminals A, B, and C. Based on a control signal from the decode control unit 706, the input terminals A,
Either B or C is selected and output. Selector 702
Are connected so that the output data of the delay circuit 705 is input to the input terminal A, the output data of the latch 701 is input to the input terminal B, and the read data from the waveform memory 8 is input to the input terminal C. Have been.

The selector 703 has two input terminals A, B
And selects and outputs one of the input terminals A and B based on a control signal from the decode control unit 706. The input terminal A of the selector 703 is connected so that the upper 8 bits of the read data from the waveform memory 8 are input, and the lower 8 bits are input to the input terminal B.

The selected outputs of the selectors 702 and 703 are input to an adder / subtractor 704. The adder / subtractor 704 performs addition or subtraction based on a control signal from the decode control unit 706. The output of the adder / subtractor 704 is input to the delay circuit 705. The output of the delay circuit 705 is input to the input terminal A of the selector 702 and also to the data selection unit 509.

The decode control section 706 is provided with selectors 702 and 703 and an adder / subtractor 70 based on the data AF1.
4 to output a control signal.

Next, the decoder section 508 will be described in detail with reference to FIG. 7 and FIGS.

As described above, the value of AF1 (0, 1,
Since the type of reading waveform data differs depending on 2), the operation of the decoder unit 508 will be described below for each value of AF1.

FIG. 8 is a time chart for explaining the operation of the decoder unit 508 when AF1 = 0. In this figure, “memory read data” indicates data sequentially read from the waveform memory 8 and input to the decoder unit 508. When AF1 = 0, the above-mentioned type 1 reading is performed, so that the linear sample L (number 801) is first compressed, and then the compressed sample one address before the linear sample L−
1D and 0D (numbering 802) are finally compressed samples -3D and -2D (numbering 803) one address before that.
Is input to the decoder unit 508.

In FIGS. 8, 9 and 10, a linear sample read from the waveform memory 8 and input to the decoder unit 508 is represented by L, and an offset value from the linear sample is added before L. Let it represent another linear sample. For example, -1L represents the linear sample of the sample point immediately before L, and + 1L represents the linear sample of the sample point immediately after L. The same applies to compressed samples. For example, 0D is a compressed sample of the upper 8 bits of the address immediately before L, -1D is a compressed sample of the lower 8 bits of the address immediately before L, and + 2D is a compressed sample of the upper 8 bits of the address next to L. Sample, + 1D is L
Represents the compressed sample of the lower 8 bits of the address next to.
The number below each block of the memory read data is
Indicates an auxiliary address when the data is read.

In FIG. 8, "selector 702" and "selector 703" indicate the selection states of selectors 702 and 703 determined based on a control signal from decode control section 706, and "decode output" indicates the decoder section. 508 shows the output. Below the selector 703, data actually output from the selector 703 is shown. Note that the selector 703 is a selector that can take a state of outputting 0 without selecting any input terminal.

Referring to FIGS. 7 and 8, first, when linear sample L is read from waveform memory 8, (80)
1) The linear sample L is latched by the latch 701 and input to the input terminal C of the selector 702. At this time, the selector 702 selects and outputs the input terminal C (811), and the selector 703 outputs 0 (8).
21) is controlled. The adder / subtractor 704 is controlled so as to subtract the output from the selector 703 from the output from the selector 702. Therefore, the linear sample L is output from the adder / subtractor 705 and the delay circuit 705
Will be entered. After a predetermined time delay by the delay circuit 705, the linear sample L is output as a decoder output.
Is input to the data selection unit 509 (831).

The output (linear sample L) of the delay circuit 705 is input to the input terminal A of the selector 702. Next, the compressed samples 0D and -1D are read from the waveform memory 8.
Is read (802). Among them, the upper 8 bits of the compressed sample 0D are input to the input terminal A of the selector 703, and the lower 8 bits of the compressed sample-1D are input to the input terminal B of the selector 703. At this time, the selector 70
2 selects and outputs the input terminal A (812), and the selector 70
3 is controlled to a state (822) for selectively outputting the input terminal A. The adder / subtractor 704 is controlled so as to subtract the output from the selector 703 from the output from the selector 702. Therefore, the result of subtracting the compressed sample 0D from the linear sample L (that is, the linear sample-
1L) is output from the adder / subtractor 705 and the delay circuit 7
05. After being delayed for a predetermined time by the delay circuit 705, the linear sample-1L is input to the data selection unit 509 as a decoder output (832).

The output (linear sample-1L) of the delay circuit 705 is input to the input terminal A of the selector 702. Next, the selector 702 is controlled to select and output the input terminal A (813), and the selector 703 is controlled to select and output the input terminal B (823). The adder / subtractor 704 is controlled so as to subtract the output from the selector 703 from the output from the selector 702. Therefore, the result of subtracting the compressed sample-1D from the linear sample-1L (that is, the linear sample-2L) is output from the adder / subtractor 705 and input to the delay circuit 705.
After being delayed for a predetermined time by the delay circuit 705, the linear sample-2L is output as a decoder output from the data selection unit 50.
9 (833).

The output (linear sample-2L) of the delay circuit 705 is input to the input terminal A of the selector 702. Next, compressed samples -2D, -2
3D is read (803). Of these, the compressed sample-2D of the upper 8 bits is input terminal A of selector 703.
And the lower 8 bits of the compressed sample-3D are the selector 70
3 to the input terminal B. At this time, the selector 702 selects and outputs the input terminal A (814), and the selector 703 selects and outputs the input terminal A (824).
Is controlled. The adder / subtractor 704 is connected to the selector 70
Control is performed so that the output from the selector 703 is subtracted from the output from the second. Therefore, the linear sample-2L
The result obtained by subtracting the compressed sample-2D from (i.e., the linear sample-3L) is output from the adder / subtractor 705 and input to the delay circuit 705. After a predetermined time delay by the delay circuit 705, the linear sample-3L is input to the data selection unit 509 as a decoder output (834).

The output (linear sample-3L) of the delay circuit 705 is input to the input terminal A of the selector 702. Next, the selector 702 is controlled to select and output the input terminal A (815), and the selector 703 is controlled to select and output the input terminal B (825). The adder / subtractor 704 is controlled so as to subtract the output from the selector 703 from the output from the selector 702. Therefore, the result obtained by subtracting the compressed sample -3D from the linear sample -3L (that is, the linear sample -4L) is output from the adder / subtractor 705 and input to the delay circuit 705.
After a predetermined time delay by the delay circuit 705, the linear sample-4L is output as a decoder output from the data selection unit 50.
9 (835).

As described above, the decoder unit 508 outputs five linear samples L, -1L, -2L, and -3.
L and -4L are sequentially output (831 to 835).

When AF1 = 1 FIG. 9 is a time chart for explaining the operation of the decoder section 508 when AF1 = 1. The operation of the decoder unit 508 in this case is the same as that in the case where AF1 = 0. However, when AF1 = 1, the above-mentioned type 2 reading is performed, so that the linear sample L (number 9
01), then the compressed sample -1D, 0D (numbering 902) one address before the linear sample, and finally the compressed sample at the next address of the linear sample +2
D and + 1D (numbering 903) are input to the decoder unit 508.

The control of the selectors 702 and 703 and the adder / subtractor 704 is also performed according to such an input. In particular, when the output from the selector 703 is a compressed sample to be added to the linear sample,
4 is controlled so that the output of the selector 702 and the output of the selector 703 are added.

For example, a compressed sample +
When 2D and + 1D are read (903), the upper 8 bits of the compressed sample + 2D are input to the input terminal A of the selector 703, and the lower 8 bits of the compressed sample + 1D are input to the input terminal B of the selector 703. At this time, the selector 702 selects and outputs the input terminal B (91
4), the selector 703 is controlled to the state (924) for selecting and outputting the input terminal B. The adder / subtractor 704 is controlled so as to add the output from the selector 702 and the output from the selector 703. Since the linear sample L of the latch 701 is output from the selector 702, the result of adding the linear sample L and the compressed sample + 1D (that is, the linear sample + 1L) is added to the adder / subtracter 70.
5 and input to the delay circuit 705. After being delayed for a predetermined time by the delay circuit 705, the linear sample + 1L is input to the data selection unit 509 as a decoder output (934).

Next, the selector 702 sets the input terminal A
(915), and the selector 703 outputs the input terminal A
Is selected (925), and the adder / subtractor 704 outputs the output from the selector 702 and the selector 7
03 is controlled to be added. As a result, the linear sample + 2L is input to the data selection unit 509 as a decoder output (935).

As a result, five linear samples L, -1L, -2L, + 1L, + 2L are output from the decoder unit 508.
Are sequentially output (931 to 935).

When AF1 = 2 FIG. 10 is a time chart for explaining the operation of the decoder unit 508 when AF1 = 2. The operation of the decoder unit 508 in this case is as described above and AF1 =
It is similar to the case of 0,1. However, when AF1 = 2,
Since the above-mentioned type 3 reading is performed, first, the linear sample L (numbering 1001), and then the next compressed sample + 2D, + 1D (numbering 1002) of the linear sample
Finally, the compressed sample at the next address +4
D and + 3D (numbering 1003) are input to the decoder unit 508.

Then, the control of the selectors 702 and 703 and the adder / subtractor 704 is also performed according to such an input. In particular, in this case, since all outputs from the selector 703 are compressed samples to be added to the linear samples, the adder / subtractor 704 is controlled to add the output of the selector 702 and the output of the selector 703.

As a result, five linear samples L, + 1L, + 2L, + 3L and + 4L are output from the decoder unit 508.
Are sequentially output (1031 to 1035).

Next, the data selection unit 509 will be described in detail.

Referring to FIG. 7, data selection unit 509 includes:
A delay circuit 711, a selector 712, and a selection control unit 713 are provided. The output from the decoder unit 508 is input to the delay circuit 711 and to the input terminal B of the selector 712. The output of the delay circuit 711 is input to the input terminal A of the selector 712. The selector 712 selects and outputs one of the input terminals A and B based on a control signal from the selection control unit 713. The selection control unit 713, based on the data AF2,
1 to output a control signal.

FIG. 11 shows a reproducing section, that is, the decoder section 5.
8 shows a time chart in the data selection unit 509 and the interpolation unit 510. Reference numeral 1101 denotes a time slot for each channel, and 1102 denotes a signal read out from the waveform memory 8 in the channel of interest and read from the decoder unit 5.
08 indicates read data to be input. R1 corresponds to read data 801 in FIG. 8, read data 901 in FIG. 9, and read data 1001 in FIG. Similarly, R
2, R3 are the read data 802, 803 of FIG.
10 and the read data 1002 and 1003 of FIG.

Reference numeral 1103 in FIG. 11 denotes a decode output which is output as a result of decoding these read data by the decoder unit 508. X1 is the decode output 831 in FIG. 8, the decode output 931 in FIG.
0 corresponds to the decoded output 1031. Similarly, X2
X3, X4 and X5 are the decode outputs 832 and 83 of FIG.
3, 834, 835, decode outputs 931 and 93 in FIG.
2, 933, 934 and the decode output 10 of FIG.
32, 1033, 1034, and 1035.

The decode output 1103 is input to the delay circuit 711 of the data selection unit 509 and the input terminal B of the selector 712 in FIG. Delay circuit 7
11 is output after a predetermined delay time, and FIG.
1 such as 1104. Selector 712 is numbered 1
One of the inputs to input terminals A and B shown at 103 and 1104 is selectively output. S1 to S of number 1105 in FIG.
4 indicates the selected output.

FIG. 11 (i) specifically shows how the data X1 to X5 of numbers 1103 and 1104 are selected by the selection control in the data selection unit 509 for each AF2. . Rectangular blocks indicate data to be selectively output, and symbols A and B above the rectangular blocks
Indicates which of the input terminals A and B the selector 712 selects and outputs.

When AF2 = 1, 2, 4, the selector 7
Reference numeral 12 is controlled so as to select and output the input terminals A, A, A, A. Therefore, X1, X
2, X3, and X4 are output in this order.

When AF2 = 3, the selector 712 is controlled to select and output the input terminals A, A, B, B. Therefore, X1, X2, X
4, and X5.

When AF2 = 0,5, the selector 712
Are controlled to select and output the input terminals B, B, B, B. Therefore, X2, X3,
X4 and X5 are output in this order.

Considering the decoded output described with reference to FIGS. 8 to 10, the linear samples sequentially output from the data selection unit 509 and input to the interpolation unit 510 are as follows. When AF2 = 0 ...- 1L, -2L, -3L,
-4L When AF2 = 1 ... L, -1L, -2L, -3
When L AF2 = 2 ... L, -1L, -2L, +1
When L AF2 = 3 ... L, -1L, + 1L, +2
When L AF2 = 4 ... L, + 1L, + 2L, +3
When L AF2 = 5 ... + 1L, + 2L, + 3L,
+ 4L

As described above, although the output order does not correspond to the temporal order of the sample points (the order is determined according to AF2), four adjacent linear samples are sequentially input to the interpolation unit 510. .

Next, the interpolation section 510 will be described in detail.

Referring to FIG. 7, interpolation section 510 includes a multiplier 721, a coefficient memory 723, an upper bit generator 722,
And an interpolation accumulator 724. Data selector 5
The output from 09 is input to the multiplier 721. Upper bit generator 722 generates the upper two bits of the read address of coefficient memory 723 based on data AF2. The coefficient memory 723 receives data AF3 (10 bits) in addition to the 2 bits from the upper bit generator 722, and reads coefficient data by using the combined 12-bit data as an address.

The four-point interpolation is obtained by multiplying each of four adjacent linear samples LS0, LS1, LS2, LS3 by a predetermined coefficient, and accumulating the multiplication results. That is, the coefficients are A0 (j), A1 (j), A2 (j),
Assuming that A3 (j), A0 (j) · LS0 + A1 (j) · LS
1 + A2 (j) · LS2 + A3 (j) · LS3, the interpolated amplitude value can be obtained.

Here, the linear samples LS0, LS1,
LS2 and LS3 are arranged in time order as sample points, and LS3 is the linear sample of the first (old) sample point, and LS1 is the linear sample of the next (new) sample point. Linear sample.

As described above, since four adjacent linear samples are sequentially input to the interpolation unit 510, if a predetermined coefficient is output from the coefficient memory 723 when each linear sample is input, The multiplier 721 multiplies each linear sample by a coefficient, and the result of the multiplication can be accumulated by the interpolation accumulator 724. This allows
The calculation of the above equation is performed, and the interpolation result is output.

The coefficients output from the coefficient memory 723 will be described in more detail.

FIG. 12 shows the contents stored in the coefficient memory 723. The horizontal axis represents the address, and the vertical axis represents the coefficient data to be read. The address input to the coefficient memory 723 is 12 bits, but the upper 2 bits (input from the upper bit generator 722) are an integer part and the lower 10 bits (data AF3).
Is regarded as a decimal part. Therefore, the horizontal axis in FIG. 12 ranges from 0 to 4. The positions of 0, 1, 3, and 4 are determined by the integer part, and the position advanced by a decimal part from the integer part as the arrow is the read position. The sample points obtained by interpolation are in the range of 1 to 2 on the horizontal axis.

Addresses 0 to 1 in the storage contents of FIG.
Corresponds to the coefficient A0 (j). j is a decimal part. The addresses 1 to 2 in FIG.
The addresses 2 to 3 correspond to the coefficient A2 (j), and the addresses 3 to 4 correspond to the coefficient A3 (j).

On the other hand, the order of the four linear samples output from the data selection unit 509 and input to the interpolation unit 510 is as follows:
Since it is determined by the value of AF2, if it is explained by that value,
It is as follows.

When AF2 = 0 ... The order of input to the interpolation unit 510 is -1L, -2L, -3L, -4L. Therefore, when represented by the above-mentioned LS symbol, LS0, LS1, LS
2, LS3. According to this, the coefficient is A0
(j), A1 (j), A2 (j), and A3 (j) in this order. Therefore, as shown in FIG.
When 2 = 0, the upper 2 bits of the address of the coefficient memory are output in the order of 0, 1, 2, 3. As a result, the coefficients are output in the order described above.

When AF2 = 1 ... The order of input to the interpolation unit 510 is L, -1L, -2L, -3L.
When represented by the symbol of LS, LS0, LS1, LS2, L
The order is S3. According to this, the coefficients are A0 (j), A1
(j), A2 (j), and A3 (j) in this order. Therefore, as shown in (ii) of FIG. 11, when AF2 = 1, the upper two bits of the address of the coefficient memory are set to 0, 1, and 2.
Output in the order of 2 and 3. As a result, the coefficients are output in the order described above.

When AF2 = 2 Since the order of input to the interpolation unit 510 is L, -1L, -2L, + 1L,
When represented by the symbol of LS, LS1, LS2, LS3, L
The order is S0. According to this, the coefficients are A1 (j), A2
(j), A3 (j) and A0 (j) should be output in this order. Therefore, as shown in FIG. 11 (ii), when AF2 = 2, the upper two bits of the address of the coefficient memory are 1, 2, 2,
Output in the order of 3,0. As a result, the coefficients are output in the order described above.

When AF2 = 3 ... The order of input to the interpolation unit 510 is L, -1L, + 1L, + 2L.
Expressed by the symbol of LS, LS2, LS3, LS1, L
The order is S0. According to this, the coefficients are A2 (j), A3
(j), A1 (j), and A0 (j) in this order. Therefore, as shown in (ii) of FIG. 11, when AF2 = 3, the upper two bits of the address of the coefficient memory are set to 2, 3,
Output in order of 1,0. As a result, the coefficients are output in the order described above.

When AF2 = 4 ... The order of input to the interpolation unit 510 is L, + 1L, + 2L, + 3L.
When represented by the symbol of LS, LS3, LS2, LS1, L
The order is S0. According to this, the coefficients are A3 (j), A2
(j), A1 (j), and A0 (j) in this order. Therefore, as shown in FIG. 11 (ii), when AF2 = 4, the upper two bits of the address of the coefficient memory are set to 3, 2,
Output in order of 1,0. As a result, the coefficients are output in the order described above.

When AF2 = 5 ... The order of input to the interpolation unit 510 is + 1L, + 2L, + 3L, + 4L. Therefore, when represented by the symbol of LS, LS3, LS2, LS
1, LS0. According to this, the coefficient is A3
(j), A2 (j), A1 (j), and A0 (j) in this order. Therefore, as shown in FIG.
When 2 = 4, the upper two bits of the address of the coefficient memory are output in the order of 3, 2, 1, 0. As a result, the coefficients are output in the order described above.

As described above, finally, 1 in FIG.
The interpolation result is output as shown at 106.

The contents stored in the coefficient memory shown in FIG.
Usually, it is symmetrical. Therefore, for example, if addresses in the range of 2 to 4 are converted so as to be folded and accessed, the coefficient memory can be halved.

According to the present embodiment, the counter 503 shown in FIG.
Is determined, one waveform amplitude value can be output by the above-described operation. Therefore, reproduction can be performed from any sample point, and skip reading of the waveform memory can be performed. Further, in the present embodiment, since data for six points can be stored in four addresses, the waveform data is compressed at a compression ratio of 4/6 = 66.7% as compared with storing one linear sample in one address. Can be stored. Furthermore, since it is stored in a compressed state, the number of times of reading (the number of times of memory access)
Requires less. Since more samples can be reproduced in a short time, there are effects such as expansion of the number of sounds.

FIG. 13 shows a modification of the format of the waveform data. FIG. 13A shows an example in which data for eight points is stored in five addresses. The data compression ratio is 5/8 =
62.5%. In this case, 4
If data for the address is read, six-point interpolation is possible. Considering the principle diagram similar to FIG. 4, there are five overlapping ranges that can be reproduced by four addresses including linear samples.

FIG. 13B shows an example in which the number of bits of the compressed sample D is 6 bits, and eight addresses are packed and stored in three addresses. If you include the linear sample,
This means that data for eight points is stored in four addresses. The data compression ratio becomes 4/8 = 50%. In this case, 1
By reading data for 4 addresses in the channel,
Seven-point interpolation is possible.

[0153]

As described above, according to the present invention,
Linear (uncompressed) amplitude value data is provided for each of a plurality of predetermined sample points, and compressed data is provided between them. Since linear amplitude value data can be reproduced in both the forward direction and the reverse direction, The waveform memory can be skipped and read without limiting the pitch width (restriction of pitch-up). Further, the waveform memory is data-compressed, and the capacity is small. Further, the number of times of reading the waveform memory can be reduced.

[Brief description of the drawings]

FIG. 1 is a block diagram of an electronic musical instrument to which a waveform generation device according to an embodiment of the present invention is applied.

FIG. 2 is a waveform data format diagram.

FIG. 3 is a diagram showing a state of generation of write data at the time of recording;

FIG. 4 is a principle diagram of reproduction of waveform data.

FIG. 5 is a block diagram of a sound source unit.

FIG. 6 is a time chart showing a change of an address at each position in the sound source section and a time chart showing an output state of an auxiliary address;

FIG. 7 is a block diagram of a reproducing unit of the sound source unit.

FIG. 8 is a time chart for explaining the operation of the decoder unit (part 1);

FIG. 9 is a time chart for explaining the operation of the decoder unit (part 2);

FIG. 10 is a time chart for explaining the operation of the decoder unit (part 3);

FIG. 11 is a time chart for explaining the operation of the reproducing unit.

FIG. 12 is a diagram showing storage contents of a coefficient memory;

FIG. 13 is a diagram showing a modification of the format of the waveform data.

[Explanation of symbols]

DESCRIPTION OF SYMBOLS 1 ... Panel switch (SW), 2 ... Keyboard, 3 ... Microcomputer (microcomputer), 4 ... Microphone, 5 ...
Analog-to-digital converter (ADC), 6 ... buffer R
AM, 7 sound source section, 8 waveform memory, 9 digital-to-analog converter (DAC), 10 sound system.

Claims (8)

(57) [Claims] An uncompressed waveform amplitude value data is provided for each of a plurality of predetermined sample points, and for a sample point between sample points of the uncompressed waveform amplitude value data, an uncompressed waveform of an immediately preceding sample point is provided. A compressed compressed waveform so that the uncompressed waveform amplitude value data can be calculated using the amplitude value data, and the uncompressed waveform amplitude value data can be calculated using the uncompressed waveform amplitude value data at the immediately following sample point. Waveform storage means provided with amplitude value data; reading means for reading one uncompressed waveform amplitude value data and a plurality of compressed waveform amplitude value data included in a continuous range from the waveform storage means; Using two uncompressed waveform amplitude value data,
Calculating means for calculating non-compressed waveform amplitude value data from the read compressed waveform amplitude value data;
As for the compressed waveform amplitude data corresponding to the sample point on the future side when viewed from the two uncompressed waveform amplitude value data, the compressed waveform amplitude values are sequentially shifted from the uncompressed waveform amplitude value data toward the future. Data is calculated, and the compressed waveform amplitude value data corresponding to the sample point on the past side temporally from the one read uncompressed waveform amplitude value data is calculated from the uncompressed waveform amplitude value data on the past side. A waveform generation apparatus comprising: a calculator for sequentially calculating compressed waveform amplitude value data in a direction toward..., And reproducing means for reproducing waveform data based on the obtained non-compressed waveform amplitude data. 2. An uncompressed waveform amplitude value data is provided for each of a plurality of predetermined sample points, and for a sample point between sample points of the uncompressed waveform amplitude value data, an uncompressed waveform of the immediately preceding sample point is provided. A compressed compressed waveform so that the uncompressed waveform amplitude value data can be calculated using the amplitude value data, and the uncompressed waveform amplitude value data can be calculated using the uncompressed waveform amplitude value data at the immediately following sample point. Waveform storage means provided with amplitude value data; frequency information generation means for generating frequency information corresponding to the pitch of a musical tone waveform to be generated; and repeatedly accumulating the frequency information at a predetermined speed to obtain an accumulated value. Accumulating means for outputting; and reading out one uncompressed waveform amplitude value data and a plurality of compressed waveform amplitude value data included in a continuous range from the waveform storage means based on the accumulated value. And means, and using a single non-compressed waveform amplitude value data was the reading,
Calculating means for calculating non-compressed waveform amplitude value data from the read compressed waveform amplitude value data;
As for the compressed waveform amplitude data corresponding to the sample point on the future side when viewed from the two uncompressed waveform amplitude value data, the compressed waveform amplitude values are sequentially shifted from the uncompressed waveform amplitude value data toward the future. Data is calculated, and the compressed waveform amplitude value data corresponding to the sample point on the past side temporally from the one read uncompressed waveform amplitude value data is calculated from the uncompressed waveform amplitude value data on the past side. And calculating means for sequentially calculating the compressed waveform amplitude value data in the direction toward, and selecting means for selecting non-compressed waveform amplitude value data to be used for interpolation from the obtained non-compressed waveform amplitude value data. A waveform generating apparatus comprising: an interpolation unit that performs an interpolation process using uncompressed waveform amplitude value data and outputs an interpolation result as waveform data. 3. The sound is sampled at a predetermined sampling frequency.
Storage device that compresses and stores waveform data obtained by
And obtaining M (M is an integer of M ≦ n / 2 + 1) uncompressed waveform amplitude value data for each of a plurality of predetermined sample points , obtaining n samples of the waveform data. For the sample points between the sample points of the waveform amplitude value data, the non-compressed waveform amplitude value data of the immediately preceding sample point can be used to calculate the non-compressed waveform amplitude value data. Compressed so that uncompressed waveform amplitude value data can be calculated using uncompressed waveform amplitude value data.
A waveform storage device provided with (n-1) pieces of compressed waveform amplitude value data. 4. An uncompressed waveform amplitude data arbitrarily selected from the uncompressed waveform amplitude data provided for each of the plurality of predetermined sample points, and using the selected uncompressed waveform amplitude data, When a corresponding plurality of uncompressed waveform amplitude value data is calculated from a plurality of continuous compressed waveform amplitude value data immediately after the uncompressed waveform amplitude value data, the last non-compressed waveform amplitude value data of the calculated plurality of uncompressed waveform amplitude value data is calculated. 4. The waveform storage device according to claim 3, wherein the compressed waveform amplitude value data is the same as the non-compressed waveform amplitude value data provided next to the selected one non-compressed waveform amplitude value data. 5. A waveform generating method for reading and reproducing waveform data from a waveform storage means, wherein the waveform storage means provides uncompressed waveform amplitude value data for each of a plurality of predetermined sample points, For the sample points between the sample points of the value data, the uncompressed waveform amplitude value data can be calculated using the uncompressed waveform amplitude value data of the immediately preceding sample point, and the uncompressed In addition to providing compressed compressed waveform amplitude value data so that uncompressed waveform amplitude value data can be calculated using the waveform amplitude value data, reading and reproduction of the waveform data from the waveform storage unit is performed by using the waveform A reading step of reading one non-compressed waveform amplitude value data and a plurality of compressed waveform amplitude value data included in a continuous range from the storage unit; Using the one uncompressed waveform amplitude value data
A calculating step of calculating uncompressed waveform amplitude value data from the read compressed waveform amplitude value data, wherein the uncompressed waveform amplitude value data corresponds to a sample point on the temporally future side when viewed from the read one uncompressed waveform amplitude value data. With regard to the compressed waveform amplitude value data, the compressed waveform amplitude value data is sequentially calculated from the uncompressed waveform amplitude value data in the direction toward the future, and temporally as viewed from the one read uncompressed waveform amplitude value data. For the compressed waveform amplitude value data corresponding to the past sample point, a calculating step of sequentially calculating the compressed waveform amplitude value data in the direction toward the past from the uncompressed waveform amplitude value data; A reproducing step of reproducing waveform data based on the waveform amplitude value data. 6. A waveform generating method for reading and reproducing waveform data from a waveform storage means, wherein the waveform storage means provides uncompressed waveform amplitude value data for each of a plurality of predetermined sample points, For the sample points between the sample points of the value data, the uncompressed waveform amplitude value data can be calculated using the uncompressed waveform amplitude value data of the immediately preceding sample point, and the uncompressed In addition to providing compressed compressed waveform amplitude data so that non-compressed waveform amplitude data can be calculated using the waveform amplitude data, reading and reproduction of the waveform data from the waveform storage means is performed by generating A frequency information generating step of generating frequency information according to the pitch of the power musical tone waveform; and a step of repeatedly accumulating the frequency information at a predetermined speed and outputting an accumulated value. A reading step of reading one non-compressed waveform amplitude value data and a plurality of compressed waveform amplitude value data included in a continuous range from the waveform storage means based on the accumulated value; Using one uncompressed waveform amplitude value data,
A calculating step of calculating uncompressed waveform amplitude value data from the read compressed waveform amplitude value data, wherein the uncompressed waveform amplitude value data corresponds to a sample point on the temporally future side when viewed from the read one uncompressed waveform amplitude value data. With regard to the compressed waveform amplitude value data, the compressed waveform amplitude value data is sequentially calculated from the uncompressed waveform amplitude value data in the direction toward the future, and temporally as viewed from the one read uncompressed waveform amplitude value data. For the compressed waveform amplitude value data corresponding to the past sample point, a calculating step of sequentially calculating the compressed waveform amplitude value data in the direction toward the past from the uncompressed waveform amplitude value data; A selecting step of selecting uncompressed waveform amplitude value data to be used for interpolation from the waveform amplitude value data; and a non-compressed waveform amplitude value data selected by the selecting step. Interpolation performs waveform generation method characterized by comprising an interpolation step of outputting the interpolation result as the waveform data with. 7. A sound sampler at a predetermined sampling frequency.
Waveform storage method for compressing and storing waveform data obtained by
A law, together with the obtaining n samples of waveform data, M (M is M ≦ n / 2 + 1 integer) for every predetermined plural sample points provided pieces of non-compressed waveform amplitude value data, uncompressed For the sample points between the sample points of the waveform amplitude value data, the non-compressed waveform amplitude value data of the immediately preceding sample point can be used to calculate the non-compressed waveform amplitude value data. Compressed so that uncompressed waveform amplitude value data can be calculated using uncompressed waveform amplitude value data.
A waveform storage method comprising providing (n-1) pieces of compressed waveform amplitude value data. 8. An uncompressed waveform amplitude value arbitrarily selected from the uncompressed waveform amplitude data provided for each of the plurality of predetermined sample points, and using the selected uncompressed waveform amplitude value data, When a corresponding plurality of uncompressed waveform amplitude value data is calculated from a plurality of continuous compressed waveform amplitude value data immediately after the uncompressed waveform amplitude value data, the last non-compressed waveform amplitude value data of the calculated plurality of uncompressed waveform amplitude value data is calculated. 8. The waveform storage method according to claim 7, wherein the compressed waveform amplitude value data is the same as the non-compressed waveform amplitude value data provided next to the selected one non-compressed waveform amplitude value data.